计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (24): 81-82.DOI: 10.3778/j.issn.1002-8331.2008.24.023

• 理论研究 • 上一篇    下一篇

新型分阶段粒子群优化算法

曾传华1,申元霞1,2   

  1. 1.重庆文理学院 数学与计算机科学系,重庆 402160
    2.西南交通大学 信息科学与技术学院,成都 610031
  • 收稿日期:2007-10-25 修回日期:2008-01-17 出版日期:2008-08-21 发布日期:2008-08-21
  • 通讯作者: 曾传华

Novel multistage Particle Swarm Optimization algorithm

ZENG Chuan-hua1,SHENG Yuan-xia1,2   

  1. 1.Department of Mathematic & Computer Science,Chongqing University of Arts and Science,Chongqing 402160,China
    2.Department of Information Science and Technology,Southwest Jiaotong University,Chengdu 610031,China
  • Received:2007-10-25 Revised:2008-01-17 Online:2008-08-21 Published:2008-08-21
  • Contact: ZENG Chuan-hua

摘要: 针对粒子群优化算法的“早熟”问题,提出了一种新型分阶段粒子群优化算法。该算法通过调整惯性权重和加速系数使粒子自组织地跟踪局部吸引域和全局吸引域来扩大粒子的搜索空间和提高粒子的收敛精度,同时根据粒子处于不同的阶段实施相应的变异策略来增加种群的多样性。通过经典函数的测试结果表明,新算法的全局搜索能力有了显著提高,并且能够有效避免早熟问题。

关键词: 粒子群优化算法, 惯性权重, 加速系数

Abstract: A novel multistage particle swarm optimization is developed for solving premature convergence of particle swarm optimization.The particles are organized to track the domain of attraction of local optimum for enlarging search space and the domain of attraction of global optimum for improving convergence performance by adaptively adjusting the acceleration coefficients and the inertia weight.Meanwhile the corresponding strategies with mutation are adopted in different stages of the new algorithm to further enhance diversity of population.Experimental results for complex function optimization show this algorithm improves the global convergence ability and efficiently prevents the algorithm from the local optimization and early maturation.

Key words: Particle Swarm Optimization(PSO), inertia weight, acceleration coefficients